Your AI writes the code.
You review it.

You're the human in the loop.

An agent-native, local-first code review tool that gives you a GitHub PR-like interface for AI-generated commits — with structured feedback your AI agent can actually act on. Works with Claude Code, Cursor, Windsurf, Codex, and OpenCode.

CorgReview mascot CorgReview diff viewer interface

Code review for AI — the way you already know how

You prompted it. Own it.

Nobody wants to review AI-generated code they didn't ask for. CorgReview lets you review your AI's output before it ever reaches your team — the same way you'd self-review before opening a PR.

The workflow you already know

Line comments, blockers, approvals, request changes. It's the PR review interface you use every day, applied to local AI commits before they go anywhere.

Feedback your AI can act on

Your review is saved as structured markdown that AI agents can read and respond to. Flag a blocker, and your agent knows exactly what to fix and where.

CorgReview brings pull request review to your local AI workflow — review it yourself first, then share with confidence.

Built into your AI coding workflow

1

Your AI agent triggers a review

When your AI agent finishes a task, it calls CorgReview via the installed skill — running corg review directly. Works with Claude Code, Cursor, Windsurf, Codex, and OpenCode. CorgReview auto-detects your feature branch and diffs it against the base. Your browser opens to a review UI showing which agent made the changes and why. No context switch, no commit ranges to remember.

Integration diagram showing Claude Code to CorgReview flow
Screenshot of the review UI with inline commenting
2

Review like a PR

Browse changed files in a resizable file tree. Read diffs with syntax highlighting — switch between unified and split view with one click. Leave line comments. Tag issues as blockers, questions, or suggestions. Set your status: approve, request changes, or comment.

3

Your AI reads the feedback and iterates

Submit your review, and it's saved as structured markdown. Your AI agent receives the results — blocker count, comments, and your summary — and addresses your feedback. When it commits fixes, re-open the same branch review and your previous comments are still there. Iterate until you approve.

Markdown file output from a review

Three ways to use it: Agent skill (corg init) for seamless AI agent integration, standalone CLI (corg review) for manual reviews, or corg serve for an interactive dashboard. Run corg init to set up in seconds.

Everything you need. Nothing you don't.

GitHub-style diff view with split comparison

GitHub-style diff view

Syntax-highlighted diffs with line numbers, file status badges, and hunk-by-hunk navigation. Toggle between unified and split view — your preference persists across sessions. Exactly what you're used to.

Line-level comments with types

Click any line to add a comment. Tag it as a blocker, question, suggestion, or note. Your AI knows which ones are mandatory.

Adding a line comment with type selection
Structured markdown output file

Structured markdown output

Every review is saved as a human-readable, git-friendly markdown file. Audit trail included.

Works with every major AI coding agent

Run corg init and CorgReview auto-detects your agent and installs natively — skills for Claude Code, Cursor, Windsurf, Codex, and OpenCode. MCP config for agents that support it, markdown-based skills for those that don't. One command, zero config.

Claude Code Cursor Windsurf Codex OpenCode
$ corg init
Detected agents:
1. Claude Code
2. Cursor
Select agent [1]: _

Branch-based reviews that pick up where you left off

CorgReview auto-detects your feature branch and diffs it against the base. Leave comments, close the browser, come back later — your review resumes exactly where you left off. Each branch gets its own persistent review file, so nothing is lost between iterations.

$ corg review
Detected branch: feat/auth-flow
Base branch: main
Resuming previous review (3 comments)...
Review UI: http://localhost:8080
CorgReview Dashboard
feat/auth-flow 3 commits ahead
fix/nav-bug 1 commit ahead
refactor/api-layer 7 commits ahead

Interactive dashboard with corg serve

Run corg serve to get a persistent dashboard. Browse branches, see recent commits, and start a review with one click — no need to remember commit ranges. The server stays alive between reviews.

Review descriptions from your AI

Your AI agent can provide a description of what the branch does and why. It shows at the top of the review UI, giving you context before you read a single line of code.

Light and dark themes

Switch between light and dark mode. Your preference persists across sessions and applies to diffs, comments, and the entire UI.

Resizable, customizable UI

Drag to resize the file tree and review summary panels. Collapse or expand the summary with one click. Switch between unified and split diff views. All your preferences persist automatically.

Runs locally, stays local

No cloud, no account, no data leaving your machine. A single binary that works offline.

Built for developers. Ready for teams.

Developers using AI coding tools

  • Review your AI's work before anyone else sees it
  • Catch issues while context is fresh — not during team code review
  • Works with Claude Code, Cursor, Copilot, or any tool that commits code
  • No workflow change — same review patterns you already use

Engineering leads & teams

  • Establish a norm: if you prompted it, you reviewed it
  • Structured review trail for every AI-generated commit
  • Visibility into what was reviewed, what was flagged, what was approved
  • Lightweight oversight without adding process

Whether you're a solo developer shipping faster with AI, or a team lead who needs to trust what's going into the codebase — CorgReview fits.

Get CorgReview

One purchase. All features. Every update through v1.x.

$19

Includes all updates through v1.x and your permanent license

  • macOS & Linux*
  • Native integration with Claude Code, Cursor, Windsurf, Codex, and OpenCode
  • Auto-detect agents with corg init — one command setup
  • All features, no restrictions
  • Every update through v1.x
  • Direct line to the developer for feedback and requests
Coming Soon

No subscription. No account required. One binary, yours forever.

*macOS users: Right-click → Open on first run (standard for unsigned binaries)